Orientation selectivity in goggle-reared kittens: An overcomplete unsupervised learning model
Gatsby Computational Neuroscience Unit, UCL , UK
The selectivities of V1 neurons are often considered to be adapted to the statistics of natural images. Accordingly, simple cell-like tuning emerges when unsupervised learning models that find sparse representations of input probabilities are trained on natural scenes. Changing the input statistics offers a stringent test of such models. We considered the case of the altered neural response properties found in goggle-reared kittens, and found that neurons in our model behaved similarly. Thus, we show that a functional model which previously has described primarily non-stimulus driven cortical organization is shown to apply also to this case of stimulus driven cortical plasticity.